The Folin–Ciocalteu method (16) was used to determine the total phenolic content and modified accordingly. A UV–Vis spectrophotometer was utilized to measure absorbance at 735 nm. The TPC used gallic acid as a standard.

The total flavonoid content was measured using the aluminum nitrate-sodium nitrite colorimetric method as described in (17) with minor adjustments. The absorbance of the samples was measured at 510 nm using a UV–Vis spectrophotometer. The TFC used rutin as a standard.

The total triterpene content was measured using the vanillin-ice acetic acid method (18). Absorbance measurements of the samples were conducted at 543 nm using a UV–visible spectrophotometer. The TTC used ursolic acid as a standard.

Ellagic acid was quantified using the Ultimate 3000 high-performance liquid chromatography system. The concentrated extract was centrifuged at 10,000 rpm for 10 min, and 2 mL of the supernatant was collected. To this, 2 mL of acidified methanol (containing 1.2 mol/L hydrochloric acid) was added, and the mass was measured. The solution was then refluxed at 85°C for 6 h, cooled to room temperature, and reweighed. The reduced weight was adjusted with methanol. The resulting solution was diluted 10 times with methanol and filtered through a 0.45 μm microporous filter membrane for ellagic acid content analysis. The chromatographic column used was PntulipsTMBP-C18 (4.6*250 mm, 5 μm), with a mobile phase of acetonitrile (A) and 0.25% formic acid in water (B). The injection volume was 10 μL, the column temperature was maintained at 35°C, and the solvent flow rate was 0.8 mL/min. The detection wavelength was set at 254 nm. The gradient elution for ellagic acid analysis was as follows: 0–2 min, 94–94% (B); 2–3 min, 94–84% (B); 3–10 min, 84–80% (B); 10–20 min, 80–94% (B).

For the determination of vitamin C content, the concentrated extract was centrifuged at 10,000 rpm for 10 min, and the supernatant was filtered through a 0.45 μm microporous filter membrane. This filtered solution was used for the measurement of vitamin C content. Chromatographic column: PntulipsTMBP-C18 (4.6*250 mm, 5 μm), mobile phase: acetonitrile (A)-0.25% formic acid in water (B). With an injection volume of 10 μL, a column temperature of 35°C, and a solvent flow rate of 0.8 mL/min, the detection wavelength: 254 nm, the gradient elution was 0–30 min, 90–80% (B).

The DPPH Free Radical Scavenging Ability Assay Kit was used to assess the samples’ ability to scavenge DPPH free radicals. The sample solution was diluted, mixed with DPPH working solution, and incubated in the dark for 30 min at room temperature before measuring absorbance at 515 nm. Trolox served as a positive control. DPPH radical scavenging rate (Ds%) was calculated as follows:

The ABTS assay kit was employed to determine the free radical scavenging ability of the samples. The sample solution was diluted, mixed with ABTS working solution and Reagent IV application solution, and incubated at room temperature for 6 min in the dark before absorbance measurement at 405 nm. Trolox was used as a positive control. The free radical scavenging rate (Ds%) of ABTS was calculated as follows:

The FRAP method was utilized to evaluate the total antioxidant capacity following the T-AOC assay kit instructions. The sample solution was diluted, mixed with distilled water and FRAP working solution, incubated for 10 min, and absorbance was measured at 593 nm.

Figure 1A shows that the total phenolic content reached its peak when the ratio of B. subtilis to S. cerevisiae was 3:1. It is important to maintain proper strain ratios during fermentation to ensure optimal microbial growth and enzyme production. Lower strain ratios may lead to insufficient growth and reduced enzyme production, while higher ratios can promote rapid biomass proliferation and enzyme synthesis (19). The activity of neutral protease produced by B. subtilis was found to increase and then decrease with the increase in inoculum amount (20). B. subtilis made the fermentation process more enzyme-producing and further promoted the release of polyphenols. Therefore, the TPC was maximum when the strain ratio was 3:1. After a certain point, enzyme production may decrease due to nutrient depletion, resulting in lower polyphenol content. Therefore, the ratio of B. subtilis to S. cerevisiae was selected as 3:1 for the response surface experiment.

The total polyphenol content was found to be highest when the liquid-to-material ratio was 2:1 and lowest at 3:1, according to Figure 1B. Further increasing the liquid-to-material ratio resulted in an increase in TPC until reaching a peak at 5:1, after which TPC decreased again. This suggests that the ratio of liquid to material can impact the growth and metabolism of bacteria during fermentation. Previous studies have shown that both too low and too high moisture content can negatively affect fermentation results (21). A moisture content that is too low may not provide enough water for B. subtilis and S. cerevisiae to thrive, while a moisture content that is too high can inhibit the growth of B. subtilis and enzyme production, ultimately leading to a decrease in TPC (22, 23). A liquid-to-material ratio of 2:1 was selected for further optimization due to its shorter drying time and higher TPC compared to a ratio of 5:1.

Figure 1C, it is observed that TPC increased from 12 to 24 h and peaked at 24 h during fermentation. However, TPC gradually decreased with prolonged fermentation time, likely due to nutrient depletion as strains require more nutrients for growth and metabolism. Phenolics may also be catabolized and metabolized as a carbon source, leading to a decrease in content. Therefore, the optimal fermentation time was determined to be 24 h.

Temperature also plays a crucial role in enzyme activity and yield. Figure 1D demonstrates that TPC initially increases and then decreases with fermentation temperature. The highest TPC was observed at a fermentation temperature of 31°C, after which TPC decreased. This decline in TPC at higher temperatures may be attributed to the impact of elevated temperature on enzyme activity and polyphenol content. Therefore, a fermentation temperature of 31°C was chosen for subsequent experiments.

Figure 1E, it is evident that the total phenol content (TPC) was highest at a 4% inoculum level, showing an increasing trend between 6 and 10% and a decreasing trend between 10 and 14%. Excessive inoculation led to rapid metabolism in the early stages of fermentation, depleting nutrients quickly and potentially producing inhibitory metabolites that hindered the growth of strains in later stages (24). Polyphenol enrichment was found to be more effective and cost-efficient at a 4% inoculum compared to a 10% inoculum, making 4% the optimal choice.

Based on the results of the single-factor test, the strain ratio (A), liquid-to-material ratio (B), fermentation time (C), and fermentation temperature (D) were selected as the influencing factors, and total phenol content was used as the response value to conduct a Box–Behnken response surface test to optimize the fermentation process of RRTP. The Box–Behnken experimental design scheme and results are shown in Table 1.

The results obtained in Table 1 were analyzed by regression using response surface software to obtain the multiple quadratic regression model equation for total phenol contents (Y):

Y = 27.86–0.1067A-0.2217B-0.0492C + 0.5425D-0.0800AB-0.3300AC-0.4000AD-0.3200BC + 0.1550BD-1.57CD-1.76A²-0.6021B²-1.97C²-2.35D². According to the analysis of variance of the model (Table 2), the p-value of the model for total phenol content was highly significant (p < 0.01) and the result of the misfit term was not significant (p > 0.05). Therefore, the model was reliable. The R² of the model was 0.9139 and adj. R² was 0.8278, indicating that the model fits well and can be used to predict the results of the RRTP fermentation process. As can be seen from the f-values, the order of effect of the factors on the total phenol content is as follows: D (fermentation temperature) > B (liquid-to-material ratio) > A (strain ratio) > C (fermentation time).

Response surface and contour plots generated from model equations visually represent the relationship between factors and response values, highlighting interaction strengths and relationships between factors (25). In contour maps, closer ellipses indicate stronger interactions, while closer circles indicate weaker interactions. The slope of the 3D response surface map indicates the magnitude of change in response values, with steeper slopes showing greater effects on response values (26). As shown in Figure 2, the steep slope of the response surface maps formed by factors AD and CD highlights the significant impact of fermentation temperature on total phenol content. The elliptical contour between Fermentation time (C) and Fermentation temperature (D) suggests an extremely significant interaction between the two factors.

After conducting a fitting using Design-Expert 13.0.1 software, the optimal fermentation process for RRTP was determined. The ideal conditions included a strain ratio (B. subtilis: S. cerevisiae) of 2.962:1, liquid-to-material ratio of 1.847:1 mL/g, fermentation time of 23.413 h, and fermentation temperature of 31.390°C. This resulted in a predicted total phenol content of 27.915 mg/g. The actual fermentation process conditions were adjusted slightly for practical reasons, with a strain ratio of 3:1, liquid-to-material ratio of 2:1 mL/g, fermentation time of 24 h, and fermentation temperature of 31°C. The total phenol content measured in three parallel tests under these conditions was 27.47 ± 0.33 mg/g, which was very close to the predicted value and validated the reliability of the regression model.

The fermentation process of microorganisms helps in releasing biologically active compounds from plants (27). The total phenolic content of RRTP increased significantly when fermented by different strains compared to the uninoculated group at various time points, with the highest TPC observed in B. subtilis: S. cerevisiae (1:1) among all groups (Figure 4A). After response surface optimization, the total phenol content of RRTP post-fermentation was notably higher than before fermentation (Figure 4A). It was observed that the fermentation by different strains led to an increase in TPC, with mixed strain fermentation proving to be more effective in enhancing the phenolic compounds content. This could be attributed to the hydrolysis of phenol compounds into simpler forms during fermentation, thereby exposing more phenolic hydroxyl groups (28). These findings align with those of Xi Hu et al. (29) who reported a significant increase in total phenol content in citrus pomace through probiotic fermentation.

The TFC of RRTP fermented by various strains was higher compared to the uninoculated group at different time points (Figure 4B). Following response surface optimization, the TFC of mixed bacteria fermentation was significantly elevated compared to pre-fermentation levels (Figure 4B), indicating an increase in TFC content due to fermentation. This increase is attributed to the release of bound flavonoids as free flavonoids (30). These results are consistent with the findings of Yan-Qiu Wang et al. (10) where the TFC of Rosa roxburghii Tratt pomace was found to increase post-fermentation.

At various time points, the total triterpene content of Rosa roxburghii Tratt pomace (RRTP) fermented by different strains was found to be higher compared to the uninoculated group (Figure 4C). Following optimization through a response surface test, the triterpene content of mixed bacterial fermentation was significantly increased compared to before fermentation (Figure 4C), indicating that fermentation could enhance the triterpene content. These results are consistent with a study by Yanfang Yan and colleagues (18), where the triterpene content was higher after fermentation of Rosa roxburghii Tratt fruit by S. cerevisiae compared to before fermentation.

Ellagic acid is a potent bioactive compound that exists as a dilactone. As depicted in Figure 4D, the ellagic acid content of RRTP fermented by different strains was higher than that of the uninoculated bacterial fermentation group, and the ellagic acid content of the optimized mixed bacterial fermentation was significantly higher than the uninoculated group, indicating that fermentation could increase the ellagic acid content. Similarly, Federica Moccia et al. (31) found high levels of ellagic acid in pomegranate wastes fermented with Aspergillus niger and S. cerevisiae. HPLC chromatograms of ellagic acid reference substance and sample are displayed in Supplementary Figure S1.

Figure 4E shows that the Vitamin C content after fermentation by different strains remained higher than that of the uninoculated bacterial fermentation group. Moreover, the optimization of mixed bacteria significantly increased the Vitamin C content, suggesting that fermentation can boost the Vitamin C content. This aligns with findings from Babatunde Stephen Oladeji and colleagues, who reported an increase in Vitamin C content following fermentation (32). The Vitamin C content increased post-fermentation but remained relatively low, likely due to the higher presence of Vitamin C in the fresh fruit of Rosa roxburghii Tratt compared to the pomace after juicing, as well as the reduction in Vitamin C content caused by drying post-fermentation. HPLC chromatograms of Vitamin C reference substance and sample are displayed in Supplementary Figure S2.